~luke-jr/moo/unknown_bifuncs : changes

~luke-jr/moo/unknown_bifuncs » Changes from revision 49

From Revision 25 to 6

Rev	Summary	Authors	Date
25	Fixed one more Log keyword that slipped by. Fixed one more Log keyword that slipped by.	nop	25 years ago
24	Merge UNSAFE_OPTS (ref fixups); fix Log tag placement to fit... Merge UNSAFE_OPTS (ref fixups); fix Log tag placement to fit CVS whims	nop	25 years ago
23	Initial string interning during db load. Initial string interning during db load.	nop	26 years ago
22	Release of r5. Release of r5.	nop	26 years ago
21	Bump up cache size to 7507, since lambdamoo has been running... Bump up cache size to 7507, since lambdamoo has been running with that for months.	nop	26 years ago
20	Merge UNSAFE_OPTS (r5) after extensive testing. Merge UNSAFE_OPTS (r5) after extensive testing.	nop	26 years ago
19	Reorganize the top of run() slightly to make it slightly mor... Reorganize the top of run() slightly to make it slightly more efficient (we do execute it billions of times, after all). Later we'll want to get rid of if (task_killed) by introducing BI_KILL or by moving it into the BI_FUNC_CALL case, at least.	bjj	27 years ago
18	1.8.0p6 merge by hand 1.8.0p6 merge by hand	nop	27 years ago
17	1.8.0p6 merge by hand. 1.8.0p6 merge by hand.	nop	27 years ago
16	A few malloc-friendly changes: rt_stacks are now centrally ... A few malloc-friendly changes: rt_stacks are now centrally allocated/freed so that we can keep a pool of them handy. rt_envs are similarly pooled. Both revert to malloc/free for large requests.	bjj	27 years ago
15	With 1.2 (oops) add MIN/MAX macros that do the obvious thing... With 1.2 (oops) add MIN/MAX macros that do the obvious thing, with undef to avoid clashing with system definitions.	bjj	27 years ago
14	* empty log message * * empty log message *	bjj	27 years ago
13	3 opcode optimizations: 3 opcode optimizations: 1) OP_IMM+OP_POP is "peephole optimized" away at runtime. This makes verbdocs and other comments cheaper. 2) OP_PUT_n+OP_POP is similarly optimized (PUT doesn't consume the top value on the stack but it is often used that way in statements like `var = expr;'). OP_G_PUT could use the same change but is rarely executed. 3) OP_PUT_n, OP_PUSH_n which used to be in an if/else in the default case are split out into 32 cases each so the compiler can optimize it for us. These ops account for a large percentage of those executed.	bjj	27 years ago
12	fastmap is mymalloc'd, so myfree it fastmap is mymalloc'd, so myfree it	bjj	27 years ago
11	sys/time.h necessary for FD_ZERO et al...maybe. sys/time.h necessary for FD_ZERO et al...maybe.	nop	27 years ago
10	clean target doesn't force reconfiguration; distclean now do... clean target doesn't force reconfiguration; distclean now does.	nop	27 years ago
9	str_dup("") now returns the same empty string to every calle... str_dup("") now returns the same empty string to every caller	bjj	27 years ago
8	new_list(0) now returns the same empty list to every caller new_list(0) now returns the same empty list to every caller	bjj	27 years ago
7	Nobody actually uses protected properties. Make IGNORE_PROP... Nobody actually uses protected properties. Make IGNORE_PROP_PROTECTED the default.	nop	27 years ago
6	Changed to ed for make depend, since ex is a cheap plastic i... Changed to ed for make depend, since ex is a cheap plastic imitation of ed on many Linux systems. Added better optimization suggestion.	nop	27 years ago